[Wien] optic sum rules
Dear Prof. Blaha I have calculated the optical properties of TiSe2 by MBJ potential with 6000, 5 and 9 kpoints which generated after scf by x kgen as discussed in the UG. As discussed in the paper of Prof. Claudia Draxl (computer physics comm. 175, 1-14 (2006)) the sum rule 1 and 2 show 'an effective number of electrons contributing to the absoption process as a function of energy'. but my calculations lead to : 6000 kpoints: sum rule 1: Int(sigma)dw 19.14 sum rule 2: Int(eloss.w)dw18.93 sum rule 3: Int(eloss/w)dw1.51 5 kpoints sum rule 1: Int(sigma)dw 19.08 sum rule 2: Int(eloss.w)dw18.94 sum rule 3: Int(eloss/w)dw1.52 9 kpoints sum rule 1: Int(sigma)dw 10.125 sum rule 2: Int(eloss.w)dw10.035 sum rule 3: Int(eloss/w)dw1.51 the results show that in 9 kpoints the value of 'sum rule 2' significantly decreases . in the UG it has mensioned that for the optic we need dense kpoints but how many? how can I check it for optic? Best Regards Ali___ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
Re: [Wien] optic sum rules
There is definitely no need to go to such a high number of k-points. In the paper of Claudia, they were looking at metallic compounds, for which we need to have a very dense k-mesh (for the intraband transitions). In your case you expect a semiconductor and thus you will have only interband transitions. Thus 6000 kpoints is already converged to my point of view. I should say that the best way to insure the convergency in the present case is to plot the epsilon2 as a function of the energy for instance and to compare this quantity with different k-meshes. In your case, try to do 300 kpoints first and then 500, 1000, 2000, 3000 kpoints ... You will certainly reach the convergency at about 1000-2000 kpoints ... Regards Xavier P.S.: The values obtained for 9 kpoints are curious but I imagine that you had a prob Le 9/30/2013 4:59 PM, ali ghafari a écrit : Dear Prof. Blaha I have calculated the optical properties of TiSe2 by MBJ potential with 6000, 5 and 9 kpoints which generated after scf by x kgen as discussed in the UG. As discussed in the paper of Prof. Claudia Draxl (computer physics comm. 175, 1-14 (2006)) the sum rule 1 and 2 show 'an effective number of electrons contributing to the absoption process as a function of energy'. but my calculations lead to : 6000 kpoints: sum rule 1: Int(sigma)dw 19.14 sum rule 2: Int(eloss.w)dw18.93 sum rule 3: Int(eloss/w)dw1.51 5 kpoints sum rule 1: Int(sigma)dw 19.08 sum rule 2: Int(eloss.w)dw18.94 sum rule 3: Int(eloss/w)dw1.52 9 kpoints sum rule 1: Int(sigma)dw 10.125 sum rule 2: Int(eloss.w)dw10.035 sum rule 3: Int(eloss/w)dw1.51 the results show that in 9 kpoints the value of 'sum rule 2' significantly decreases . in the UG it has mensioned that for the optic we need dense kpoints but how many? how can I check it for optic? Best Regards Ali ___ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html ___ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
Re: [Wien] optic sum rules
Dear Xavier thank you very much for your reply. Actually, I have repeated the calculations again. I see the spectra are almost same for 2000 and 3000 kpoints but still there is a significant change in the spectra for 9 kpoints. how can I explain it? furthermore, when you said 2000, 3000 kpoints, your means is in the BZ or IBZ? Best Regards Ali From: Rocquefelte To: A Mailing list for WIEN2k users Sent: Monday, September 30, 2013 6:43 PM Subject: Re: [Wien] optic sum rules There is definitely no need to go to such a high number of k-points. In the paper of Claudia, they were looking at metallic compounds, for which we need to have a very dense k-mesh (for the intraband transitions). In your case you expect a semiconductor and thus you will have only interband transitions. Thus 6000 kpoints is already converged to my point of view. I should say that the best way to insure the convergency in the present case is to plot the epsilon2 as a function of the energy for instance and to compare this quantity with different k-meshes. In your case, try to do 300 kpoints first and then 500, 1000, 2000, 3000 kpoints ... You will certainly reach the convergency at about 1000-2000 kpoints ... Regards Xavier P.S.: The values obtained for 9 kpoints are curious but I imagine that you had a prob Le 9/30/2013 4:59 PM, ali ghafari a écrit : Dear Prof. Blaha > > >I have calculated the optical properties of TiSe2 by MBJ potential with 6000, >5 and 9 kpoints which generated after scf by x kgen as discussed in >the UG. >As discussed in the paper of Prof. Claudia Draxl (computer physics comm. 175, 1-14 (2006)) the sum rule 1 and 2 show 'an effective number of electrons contributing to the absoption process as a function of energy'. >but my calculations lead to : >6000 kpoints: > >sum rule 1: Int(sigma)dw 19.14 sum rule 2: Int(eloss.w)dw18.93 sum rule 3: Int(eloss/w)dw1.51 5 kpoints > >sum rule 1: Int(sigma)dw 19.08 sum rule 2: Int(eloss.w)dw18.94 sum rule 3: Int(eloss/w)dw1.52 >9 kpoints sum rule 1: Int(sigma)dw 10.125 sum rule 2: Int(eloss.w)dw10.035 sum rule 3: Int(eloss/w)dw1.51 the results show that in 9 kpoints the value of 'sum rule 2' significantly decreases . in the UG it has mensioned that for the optic we need dense kpoints but how many? how can I check it for optic? >Best Regards > >Ali > > >___ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html ___ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html___ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
Re: [Wien] optic sum rules
On 10/01/2013 11:24 AM, ali ghafari wrote: Actually, I have repeated the calculations again. I see the spectra are almost same for 2000 and 3000 kpoints but still there is a significant change in the spectra for 9 kpoints. As Xavier indicated in his PS, it almost certainly means that something is going wrong, though I cannot tell you what it is. If you see no difference in epsilon between 2k and 3k k-points, I would say that means the mesh is converged for optics. Even if you want to be very careful in your convergence test, keep in mind that all other optical quantities are calculated directly from epsilon. The only other thing you could check is convergence in the energy range you calculate Im(eps) in, because the Kramers-Kronig needs it (in principle) from 0 to infinity. how can I explain it? furthermore, when you said 2000, 3000 kpoints, your means is in the BZ or IBZ? In the full Brillouin zone. Elias ___ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
